An innovative electronic system for testing the performance of wind turbines is ... load in order to obtain a large number of steady state experimental points ...
Automatic System for Wind Turbine Testing S. M. Camporeale Dipartimento di Meccanica e Materiali, Universita` di Reggio Calabria, Italy
B. Fortunato Dipartimento di Ingegneria Meccanica e Gestionale, Politecnico di Bari, Italy
G. Marilli Alcatel, Italy
An innovative electronic system for testing the performance of wind turbines is presented. The main goal of the system is to increase the accuracy in the measurements of torque and speed for each steady-state point of the turbine characteristic power curve. Another useful advantage provided by the electronic control is given by the possibility of fine tuning the load in order to obtain a large number of steady state experimental points describing the characteristic curve of the turbine. Moreover, the system is suitable for integration into an automatic data acquisition and control system. In the paper the main characteristics of the electronic system are described and compared with a traditional system. This electronic control system is used for testing a small Vertical Axis Wind Turbine in a wind tunnel. The wind turbine is directly coupled to a direct current electric generator, and a chopper, electronically controlled by means of a Pulse Width Modulator, is used to supply the circuit. The electric generator is used for braking the wind turbine at various speeds during the performance test. The experimental results obtained through the proposed system are presented and discussed.
Introduction In order to experimentally determine the performance of a wind turbine model, it is necessary to determine the efficiency of the individual components of the wind system. For this purpose, the aerodynamic behavior of the wind turbine should be investigated independent of either the mechanical train system or the electric generator. The wind turbine can be tested in a test bed placed in wind tunnels, which provide controlled and uniform wind velocity conditions and can be specifically instrumented for research. Experimental tests in a wind tunnel are carried out in various laboratories in order to find, e.g., the best working conditions of a wind turbine 关1兴, to analyze the effect of the operating conditions on the blade damage 关2兴, to develop improved engineering models of the wind turbine 关3兴, or to examine the velocity field around the turbine to validate CFD simulation developed for the prediction of the wind turbine performance 关4,5兴. For varying the rotational speed of the turbine model not connected to the grid, a variable electric load is needed, generally realized through a variable resistor bank. The load must be varied by steps from the cut-in conditions to the maximum power and then to the cut-off conditions. The larger the number of the experimentally-determined working points, the more carefully the characteristic curve of the turbine model on the 关 C p ⫺ 兴 plane will be described. Generally the resistive load is varied using contactors able to switch on or off part of the bank 共see, e.g., Refs. 关1兴, 关2兴, 关6兴兲. The number of working points is dependent on the number and arrangement of resistors which form the resistive bank 关1兴. Moreover, during the test, the thermal drift of the resistance must be avoided by using resistors independent of temperature, in order to maintain a constant rotational speed of the wind turbine. In a previous work 关6兴, experiments on small wind turbine models, carried out by using rheostats for varying the electric load, have shown that it is difficult to obtain a sufficient number of steady points to describe the wind turbine characteristic power curve because it is difficult to have a fine regulation of the electric load. In the present paper, the authors try to overcome this problem by means of a power electronic device which permits the variation of the electric load in a continuous way. The system is based on the use of a chopper and a Pulse Width Modulator 共PWM兲 for the
control of the load. The system has been applied to the experimental investigation of the performance of a vertical axis wind turbine 共VAWT兲 in a small subsonic wind tunnel. The most innovative part of the testing equipment is represented by the regulation circuit, which is completely computer managed and allows one to determine many more data points than a traditional system based on variable resistors. The experimental test is completely monitored by means of a personal computer 共PC兲. A data acquisition system and some suitable transducers are also connected to the PC.
Regulation System Characteristics The vertical axis wind turbine is not self-starting due to the blade stall at start-up and low rotational speed. Therefore, an electric motor is needed in order to start the turbine and bring it up to a rotational speed which allows the blades to work out of the stall condition. After this transient, the turbine, moved by the wind energy, is able to supply power if the electric generator is properly connected to the electric load. By changing the electrical load, it is possible to reach different steady-state points, defining the characteristic power curve 关 C p ⫺ 兴 of the wind turbine. For this work, the wind turbine is directly connected to a permanent magnet dc electric machine, which is used both as a motor and as an electric generator. The governing equations for the dc machine 关7兴 are given by e a ⫽K m
(1)
T⫽K T i a
(2)
where K m and K T are two characteristic constants for the electric machine. Let us describe first the method used to vary the load using a variable resistance load. Figure 1 shows the equivalent circuit of the electric scheme adopted in this case. The turbine start-up is begun with the switch in position 共1兲. In this way, the dc electric drive, working as a motor, is connected to the battery through the rheostat, R s , which limits the circulating current. With the switch kept in position 共1兲, the turbine rotor is accelerated until the prescribed speed is reached. Then, with the switch turned to position 共2兲, the turbine is braked by connecting the electric drive, which becomes a generator, to the load rheostat, R L . In this case, the power supplied by the wind turbine is dissipated in the resistors of the load, R L and of the internal armature, R a . In steady-state operation, the internal torque of the electric machine is given by:
Fig. 3 Wind turbine regulation with variation of the voltage supplied to the dc machine terminals Fig. 1 Start-up and regulation system of the wind turbine by means of a dc machine and rheostats
T⫽
K TK m R L ⫹R a
(3)
In the plane of torque and rotational speed (T⫺ ), this equation is represented by a straight line whose slope changes with load resistance R L 共Fig. 2兲. Neglecting the effects of the viscous and coulomb friction for the steady-state mechanical equilibrium of the shaft, the torque of the wind turbine, T W , must be equal to the internal torque of the generator, T. Therefore, by varying the load resistance, R L , it is possible to vary the working point, obtaining the turbine performance curve. This procedure shows the following problems: • the load resistance variation is unlikely to be suitable for an automatic acquisition system, because it can only be done either by moving the rheostat slip or by switching the contactors used to connect a discrete-stepped resistance load bank 关1,2兴; • in order to describe each characteristic curve with a sufficient number of points, the adjustable load must be realized either by a rheostat with a very fine resistance variation or by a resistance bank composed of many sets of resistors to be connected together in order to obtain the desired load; • thermal drifts influence the stability of the steady-state working points, unless rheostats or resistors independent of temperature are used. In order to overcome these problems, a power electronic system able to provide a continuous load variation has been developed and built. In the proposed system 共Fig. 3兲, the dc electric drive terminals are connected to a variable voltage source, V t . The turbine can be started by gradually increasing the voltage V t . After the turbine has reached a rotational speed sufficient for producing power, the transition to braking is accomplished by decreasing the source voltage below the armature emf, e a . The armature current, i a , reverses direction, and therefore, the electric
Fig. 2 Speed control by variation of load resistance
machine becomes a generator 共dynamo兲. The internal torque of the dynamo is related to the voltage V t and to the turbine speed, , by T⫽
KT 共 K ⫺V t 兲 . Ra m
(4)
In the (T⫺ ) plane, this equation is represented by a straight line with positive slope 共Fig. 4兲; increasing the source voltage V t causes the slope to remain constant and the intercept on the torque axis to decrease. It appears that the turbine steady state working point can be controlled by varying the voltage applied to the armature terminals. An electronic regulation circuit capable of supplying a variable voltage to the dc motor has been designed and built. This circuit is composed of two fundamental parts: a two-quadrant chopper and a PWM. The chopper is a well-known dc-to-dc converter that generates a variable voltage and current from a fixed dc source. In the application described here, a two-quadrant 共Class C兲 chopper, able to provide a smooth transition from motoring to braking, is used. The basic circuit of a two-quadrant chopper is sketched in Fig. 5; the circuit utilizes two bipolar transistors 共Q 1 and Q 2 兲, which work as static switches, and two diodes 共D 1 and D 2 兲, which allow the discharge of the current circulating in the dc motor inductive load. The two transistors are alternatively switched on by signals S 1 and S 2 , so that the current circulates in the dc machine armature circuit as shown in Fig. 6 关8兴. Since the frequency of signals S 1 and S 2 is constant, the dc electric drive is supplied by a square ¯ apwave voltage at the same frequency. The average voltage V t plied to the dc electric drive is related to the battery voltage, E, by the relationship
Fig. 4 Speed control by variation of the voltage V t
Fig. 7 Block diagram of the PWM
Fig. 5 Two-quadrant chopper connected to the equivalent circuit of the dc electric machine
¯V ⫽ t ON E t tp
(5)
where t ON is the conduction period and t p represents the wave period. From the chopper theory, the average value of the armature current is ¯ ¯i ⫽ V t ⫺e a a Ra
(6)
Since the torque T is related to the armature current from Eq. 共2兲, it appears that the average value of the internal torque can be changed by means of the variation of the period t ON . Owing to the large mechanical inertia of the turbine coupled to the dc machine, the pulsating current produced by the chopper does not produce oscillations of the shaft rotational speed. Signals S 1 and S 2 are provided by a PWM circuit described by the block diagram shown in Fig. 7; the modulating input signal is compared with a triangular carrier, generated by the oscillator, by means of two open-loop operational amplifiers 关9兴. The input and the output wave-forms are shown in Fig. 8, where the signals S 1 and S 2 , suitable to drive the chopper, are plotted. It can be observed that the average value of the square wave S 1 is proportional to the voltage of the input modulating signal. In order to accomplish the regulation with a PC, the modulating signal is taken from an 8-bit digital-analog converter which is directly connected to the PC parallel port.
Fig. 8 Input and output wave-forms of the PWM circuit
It is possible to show 关9兴 that, using the above described PWM technique, the driving-voltage V t applied to the dc machine connected to the wind turbine is related to the numerical value b at the PC parallel port, by the equation ¯V ⫽ b E t 256
(7)
where b is the 8-bit string decimal value, variable from 0 to 255. This means that the regulation circuit can provide 256 different load values to the turbine under test. In the tests described in 关1兴 and 关2兴 共the traditional method based on variable resistance兲, in order to obtain the same number of electric load values, a load bank composed of eight separate banks of resistors, each with its own contactors, was adopted. In those tests, the resistance values were such that, at constant generator voltage, the next bank would sink double the current 共and power兲 of the previous bank. With this binary relation, it was possible to obtain 256 different loads, equally spaced in power. Therefore, it appears that the proposed regulation circuit based on the electronically controlled chopper allows one to obtain a careful characterization of the turbine more easily than possible with rheostats or discrete-stepped resistor banks.
Instrumentation System
Fig. 6 Signals driving the chopper „i S 1 and i S 2 … and current circulating in the dc machine „ i a …
All the experimental measurements have been obtained with suitable transducers and data acquisition cards connected to the PC. The turbine rotational speed has been measured with an electromagnetic sensor 共pick-up兲, placed near a 60-tooth gear, mounted on the shaft. The output signal is converted to a TTLcompatible signal by means of a zero-cross switch circuit. The
Fig. 9 Block diagram of the data acquisition and control system
signal frequency is evaluated by pulse counting and measuring the time required for 1000 complete cycles 共about 1 to 2 seconds兲. Since the frequency of the counter internal clock is 1 MHz, assuming that all pulses are counted, the error in the turbine speed can be considered negligible. For the torque measurement, the dc electric machine is mounted in the cradling case manner and the reaction force acting on the arm connected to the cradling case of the electric machine is measured by a dynamometer utilizing a Linear Variable Differential
Transformer 共LVDT兲 transducer for measuring the displacement of the arm 关10兴. The scales used to measure the force are calibrated using deadweights. The PC used for data acquisition has a 16-channel analog-input card and a two-channel counter card. Fig. 9 shows the block scheme of the regulation and data acquisition system. An openloop control system is adopted. The variation of the working point is obtained by varying the driving voltage via software. The wind speed measurement upstream of the rotor has been carried out using a Pitot tube connected to a differential micromanometer. The measurements have been performed at the nodes of an ideal 10-cm mesh grid, placed 1m upstream of the turbine. For every steady-state point, 100 samples of torque and speed values have been acquired in approximately 180 seconds. The braking operation is carried out by the operator, who selects, via software, the driving voltage value. Since the VAWT is not self starting, it is necessary to drive the turbine rotor up to operating speed with an electric motor. The sequence of an experimental test is the following: • the turbine model, located in the test section of the wind tunnel, is started by an electric motor; • the turbine rotor speed increases due to the wind energy until it reaches a steady state; • the required measurements are carried out a predefined number of times for each experimental point; • the data are stored and processed. Figure 10 shows a simplified flow chart of the acquisition and regulation management software which performs the operations necessary to test the turbine.
Test Results
Fig. 10 Flow chart of the wind turbine test procedure
A vertical axis turbine model with four straight blades, each inclined with respect to the rotational axis, has been tested 共Fig. 11兲. The turbine height is 0.7 m and the maximum radius is 0.37 m. The 60-cm-long blades have a NACA 0015 profile, with constant chord and thickness of 60 mm and 15 mm, respectively. The Reynolds number, referred to the profile chord at the maximum radius and rotational speed of 700 rpm, is about 120,000. The wind tunnel test section is 1⫻1 m. The dc electric drive is directly connected to the turbine shaft and is mounted outside the wind tunnel in order to avoid disturbing the air flow. Figure 12 shows the mean values of rotational speed and torque as functions of the driving voltage for a wind speed of 10 m/s. For each value of the driving voltage, the mean values of rotational speed and torque are evaluated on 100 acquired samples. For each point, a confidence range equal to the evaluated standard deviation
Fig. 13 Measured power versus rotational speed for the tested turbine
Fig. 11 Turbine under test in the wind tunnel
is also indicated. From this figure, it appears that the standard deviation of measured rotational speed is extremely small; the torque standard deviation is much larger, since it is influenced by the accuracy of the force measurements and by the less than perfect dynamic balance of the turbine rotor. The minimum torque of 0.015 Nm corresponded to the minimum force measurable by the dynamometer. The power obtained from the mean values of torque and rotation speed is shown in Fig. 13 as a function of the rotational speed. Many steady-state points have been obtained, notwithstanding the very low power produced by the tested turbine. Theo-
retically, many more points could have been taken by means of the fine load variation provided by the system, but this number of acquired steady-state points was considered sufficient. The low values of power 共corresponding to a maximum power coefficient, referred to the cross-sectional area, of about 0.016兲 are mainly due to the low Reynolds Number, deriving from the small size of the tested turbine, and due to the low wind speed inside the tunnel. Moreover, owing to the small test section area of the wind tunnel available for tests, the power coefficient values are not meaningful due to the blockage of the wake expansion downstream of the wind turbine. However, for the aims of this work, the small standard deviation range calculated for the measured quantities and the accurate characterization of the wind turbine show the validity of the proposed testing system. It appears to be characterized by high reliability and flexibility and seems to be suitable for application for testing small wind turbines either in the wind tunnel or in the open field.
Conclusions In the present paper, a system for wind turbine testing has been proposed. A VAWT directly coupled to a dc electric drive has been tested and a two-quadrant chopper electronically controlled by a PWM has been used to control the electric drive. The drive is used for motoring the turbine during the start-up and for braking during the performance test. The system is completely managed via computer and allows one to measure many turbine operating points more easily than a traditional system based on variable resistors. The system has been tested in open-loop control operation to determine stable operating points for velocity and torque at constant wind speed. The system can be considered for the design of a closed-loop control system, either for constant- or variablespeed turbine studies, with variable wind speed. The system is very simple and cheap and can be applied also to stand-alone testing applications for small turbines.
Acknowledgments Fig. 12 Average values and standard deviations of torque and rotational speed
The present research has been supported by the Italian Agency MURST 共Ministero della Ricerca Scientifica e Tecnologica兲.
Nomenclature A Cp dc ea
⫽ ⫽ ⫽ ⫽
E ia Km KT M n P r Ra RL Rs t t ON tp T U⬁ VAWT Vt
⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽ ⫽
rotor cross sectional area turbine power coefficient⫽P/(1/2 AU ⬁3 ) direct current electro-motive force 共emf兲 induced in the armature winding battery voltage current in the armature winding magnetic constant torque constant dc electric machine rotational speed 共rpm兲 turbine power output maximum rotor radius armature resistance load resistance starting resistance time chopper conduction time chopper pulsing period torque wind velocity vertical axis wind turbine voltage applied to the dc machine terminals turbine rotor tip speed ratio⫽r /U ⬁ air density
⫽ turbine rotor angular speed
References 关1兴 Carlin, P. W., and Fingersh, L. J., 1997, ‘‘Some Preliminary Results from the NTWC Direct-Drive, Variable-Speed Test Bed,’’ A Collection of the 1997 ASME Wind Energy Symposium Technical Papers Reno, Nevada, pp. 301– 308. 关2兴 Sutherland, H. J., and Carlin, P. W., 1998, ‘‘Damage Measurements on the NWTC Direct-Drive, Variable-Speed Test Bed,’’ A Collection of the 1998 ASME Wind Energy Symposium Technical Papers, Reno, Nevada, pp. 315– 322. 关3兴 Schepers, J. G., 1999, ‘‘An Engineering Model for Yawed Conditions, Developed on the Basis of Wind Tunnel Measurements,’’ A Collection of the 1999 ASME Wind Energy Symposium Technical Papers, Reno, Nevada, pp. 164 – 174. 关4兴 Fortunato, B., Dadone, A., and Trifoni, V., 1995, ‘‘A Two Dimensional Methodology to Predict Vertical Axis Wind Turbines Performance,’’ ASME J. Sol. Energy Eng., 117, No. 3, pp. 187–193. 关5兴 Grant, I., Mo, M., Pan, X., Parkin, P., Powell, J., Reinecke, H., Shuang, K., Coton, F., and Lee, D., 2000, ‘‘An experimental and numerical study of the vortex filaments in the wake of an operational, horizontal-axis, wind turbine,’’ J. Wind. Eng. Ind. Aerodyn., 85, No. 2, pp. 177–189. 关6兴 Casalini, F., and Fortunato, B., 1992, ‘‘Analisi Sperimentale in Galleria del Vento di un modello di Turbina Eolica,’’ Proc. of 47° Congresso Nazionale ATI, Parma, Italy, pp. 819– 832. 关7兴 Fitzgerald, A. E., Kingsley Jr., C., and Kusko, A., 1990, Electric Machinery, McGraw-Hill, New York. 关8兴 Dewan, S. B., and Straughen, A., 1975, Power Semiconductor Circuits, John Wiley & Sons, New York. 关9兴 Millman, J., and Grabel, A., 1992, Microelectronics, McGraw-Hill, New York. 关10兴 Doebelin, O. E., 1990, Measurement Systems, McGraw-Hill, New York.
